April 23, 2026 Webinar
SIGNET: A Signed Network Spectral Clustering Method for Proteomic Module Discovery in Alzheimer's Disease.
Panpan Zhang, PhD
Abstract:
Signed networks arise naturally in many biomedical applications, where both positive and negative relationships contain distinct and important biological information. However, most existing community detection methods either ignore edge signs or rely on transformations that do not adequately preserve negative correlation interpretation, potentially obscuring meaningful structure. In this paper, we propose SIGNET, a spectral clustering framework for signed, weighted networks that directly incorporates both positive and negative edge information through the signed Laplacian. We further develop silhouette score-based strategies for selecting the number of clusters, including a data-driven silhouette-gap criterion. Through extensive simulations, we demonstrate that SIGNET achieves robust and accurate clustering performance across a range of network configurations, including varying signal strengths, cluster sizes, and structural heterogeneity. Compared with existing approaches, SIGNET exhibits improved stability in cluster number selection and more reliable recovery of underlying modular structure. We further apply SIGNET to a protein-protein co-expression network from the Vanderbilt Memory and Aging Project. The identified modules are biologically coherent and capture biological processes relevant to Alzheimer's disease, highlighting the practical utility of SIGNET in high-dimensional biomedical studies. An R package, SIGNET, is publicly available to support implementation.
Short Bio:
Panpan Zhang is an Assistant Professor of Biostatistics at Vanderbilt University Medical Center and co-leader of the Data Management and Statistics (DMS) Core of the Vanderbilt Alzheimer's Disease Research Center (ADRC). His research focuses on developing rigorous statistical and machine learning methods for high-dimensional, multimodal biomedical data arising from multi-center cohort studies, with a particular emphasis on Alzheimer's disease and related dementias (ADRD). In addition to his research, Dr. Zhang serves as Program Chair-Elect of the ASA Statistics and Data Science in Aging (SDSA) Interest Group and as an early-career representative on the Data Core Steering Committee of the National Alzheimer's Coordinating Center (NACC).
Hope to see you at the webinar!